Announcing u2-mcp: Connect AI Assistants to Universe/UniData Databases

[Peter Amiri]

I'm excited to share an open-source project that bridges the gap between modern AI and MultiValue databases.

u2-mcp is a Model Context Protocol (MCP) server that enables AI assistants like Claude to directly query and interact with Rocket Universe and UniData databases—while preserving native MultiValue semantics.

🔑 Key Features:

• Query files using RetrieVe/UniQuery syntax

• Read/write records with full dynamic array support

• Access dictionaries and call BASIC subroutines

• Built-in safety controls (read-only mode, command blocklist, query limits)

• OAuth authentication for secure cloud deployment

💡 Why This Matters:

Instead of flattening MultiValue data into relational models, u2-mcp preserves attributes, multivalues, and subvalues natively. Your AI assistant understands your data the way your applications do.

Use Cases:

• Natural language queries against your ERP/business data

• AI-assisted report generation

• Database documentation and exploration

• Training new developers on legacy systems

The project uses Rocket's official uopy package and is available now on PyPI and GitHub.

📦 pip install u2-mcp

🔗 GitHub: github.com/bpamiri/u2-mcp

I'd love to hear from others in the MultiValue community. What would you want an AI assistant to help with in your U2 environment?